Protection System of Communication Line based on Distributed Interference Fiber Sensing Network

ABSTRACT

On the issue of security and protection of communication line, this study designed and realized a new distributed fiber vibration sensing and warning system for hazardous events. Based on M-Z interference principle, the system realized the real-time detecting and accurate locating of the vibration signals by means of cross-correlation calculation of the fiber signals, restrained effectively the sensing signal noise and extracted accurately the vibration signals of various hazardous events with the methods of wavelet threshold shrinkage and Hilbert transform. In the meantime, the study proposed multi-dimensional features including wavelet decomposition, MFCC coefficients and time-domain envelope to describe the vibration signals and established multi-level classifier models with BP neural network and SVM methods to realize the classification and recognition of various hazardous events. After the applications and experiments of the system on 7 hazardous events, we obtained experimental results of the locating accuracy of <100 meters and classifying accuracy rate of 94.35% on hazardous events, which verified the effectiveness of the study.

Keywords:

Artificial neural network, Digital signal processing, Fiber sensing network, Security of communication line.